Related Experiment Video
Updated: Mar 25, 2026
05:47
Simulation of a Scaled Assembly Process with Collaboration of a Robotic Arm and Monitoring through a Vision System for Quality Control
Published on: August 29, 2025
580
Digital-Twin-Guided Protocol for Supply Chain Risk Prediction and Mitigation with Neural Sequence and Graph Models
1Department of Journalism and Communication, Fuyang Normal University; cxyan_xyc@163.com.
Journal of Visualized Experiments : Jove
|March 23, 2026
Summary
This study introduces a digital-twin workflow for real-time supply chain risk prediction and mitigation. It enhances forecasting accuracy and enables cost-effective, proactive responses to disruptions.
Area of Science:
- Operations Research
- Supply Chain Management
- Artificial Intelligence
Background:
- Supply chains face significant risks due to high uncertainty and network complexity.
- Real-time risk prediction and mitigation are crucial for operational resilience.
- Existing methods often lack integrated prediction and proactive mitigation capabilities.
Purpose of the Study:
- To develop a reproducible digital-twin (DT) workflow for integrated supply chain risk management.
- To enhance real-time risk forecasting and enable optimized mitigation strategies.
- To provide a scalable and adaptable solution for diverse supply chain settings.
Main Methods:
- Developed a containerized digital-twin workflow integrating data ingestion, spatiotemporal risk forecasting, and mitigation policy optimization.
- Employed an interpretable sequence-and-graph model (Bi-LSTM with graph convolution) for prediction.
- Utilized proximal policy optimization (PPO) for learning cost-aware mitigation policies from simulated disruptions.
- Established a hierarchical key-risk-indicator taxonomy and provided version-pinned environments for reproducibility.
Main Results:
- Demonstrated significant improvements in predictive accuracy (e.g., 19.3% F1 gain on an automotive network) and extended early-warning horizons (5.2 hours).
- Achieved substantial cost reductions (over 20% in delays and emergency response costs) in a port congestion scenario using PPO.
- Validated the workflow's scalability to 500-node simulations with millisecond latency and stable long-horizon performance.
- Confirmed improved accuracy, robust decision quality, and practical scalability through dual validation methods.
Conclusions:
- The presented digital-twin workflow offers a robust and scalable solution for real-time supply chain risk prediction and mitigation.
- The integration of advanced AI models (Bi-LSTM, GCN, PPO) enables proactive, cost-effective responses to supply chain disruptions.
- The workflow's reproducibility features and adaptability guidelines facilitate its transfer and application to various supply chain contexts.